CN108429259B - A kind of online dynamic decision method and system of unit recovery - Google Patents
A kind of online dynamic decision method and system of unit recovery Download PDFInfo
- Publication number
- CN108429259B CN108429259B CN201810271593.3A CN201810271593A CN108429259B CN 108429259 B CN108429259 B CN 108429259B CN 201810271593 A CN201810271593 A CN 201810271593A CN 108429259 B CN108429259 B CN 108429259B
- Authority
- CN
- China
- Prior art keywords
- unit
- node
- restored
- search
- decision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/04—Circuit arrangements for ac mains or ac distribution networks for connecting networks of the same frequency but supplied from different sources
- H02J3/06—Controlling transfer of power between connected networks; Controlling sharing of load between connected networks
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/26—Pc applications
- G05B2219/2639—Energy management, use maximum of cheap power, keep peak load low
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S40/00—Systems for electrical power generation, transmission, distribution or end-user application management characterised by the use of communication or information technologies, or communication or information technology specific aspects supporting them
- Y04S40/20—Information technology specific aspects, e.g. CAD, simulation, modelling, system security
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Automation & Control Theory (AREA)
- Probability & Statistics with Applications (AREA)
- Pure & Applied Mathematics (AREA)
- Algebra (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of units to restore online dynamic decision method and system, comprising: determines black starting-up power supply when system is restored, and chooses the unit for needing preferentially to start from all units as unit set to be restored;Generate the tape label training set for covering each state in unit recovery process as far as possible.Construct valuation network;The scope of power outage of judgement system identifies that the availability of various equipment in judgement system obtains machine unit characteristic to the state of system;Using valuation network, it is search target with unit total generating capacity, carries out the search of Monte Carlo tree.Using the method for parallel computation, constraint checking is carried out to each alternative route;Summarize Monte Carlo tree search result, chooses the transmission line of electricity that will restore in next step and investment.Restore real time data according to system, the transmission line of electricity for needing to put into online dynamic decision unit recovery process instructs the unit recovery of power system recovery early period.
Description
Technical field
The present invention relates to a kind of units to restore online dynamic decision method and system.
Background technique
With the development of the social economy, electricity consumption is rapidly promoted, the load of power grid and installed capacity are continuously increased, power train
The grid structure of system and dynamic behaviour also become increasingly complicated, substantially increase the complexity of operation of power networks and maintenance.If office
Portion's troubleshooting is improper, easily causes accident range to expand, and then lead to the generation of large area blackout.Such as Ohio, USA
Short trouble occurs for 345kV transmission line of electricity touching tree in one, state, since the troubleshooting is improper, causes trend that a wide range of turn occurs
It moves, leads to a plurality of transmission line of electricity cascading trip due to overload, finally cause the U.S., North America has occurred in history most sternly in Canada
The power outage of weight causes to lose 61.8GW load altogether, affects the normal work and life of nearly 50,000,000 people;In July, 2012
30 days and 31 days, north India and eastern region recurred large area blackout twice, covered the territory of more than half,
Directly affect the life of more than 600,000,000 people.The operating experience of domestic and international electric system shows new technology and new equipment in the power system
Although extensive application can be improved system operation stability and reliability, be still unavoidable from the generation having a power failure on a large scale,
Especially manpower can not resist caused by factor and have a power failure on a large scale.
Present power supply plays a very important role in social production and life, and large-scale blackout can be to entire society
It can produce and cause adverse effect very serious with people's lives, or even threaten national security.In recent years, fairly large several times
Power outage has also beaten alarm bell for us.Since 1998, each provincial electric power company, China started to formulate respective black starting-up
Scheme.The professional standard " power system security stability control techniques directive/guide " that State Grid Corporation of China, China promulgated in 2000 is newly-increased
" restoring control " chapters and sections.In " the country's disposition electric grid large area power cut event emergency preplan " that State Council promulgated in 2005,
Specific requirement has been made to the formulation of special project emergency preplan when grid disconnection, electric grid large area power cut occurs.It further repairs within 2015
It orders and has printed and distributed " national large-area power-cuts event emergency preplan ".In order to draw the experience and lessons that India has a power failure on a large scale, Electricity Monitoring Commission, China
It has been issued again " about the opinion for reinforcing electric power safety work prevention electric grid large area power cut ", in the middle to reinforcing power emergency management
And the related deployment after having a power failure on a large scale is elaborated.
Unit recovery is basis and the guarantee of entire power system recovery, under the premise of meeting all kinds of constraints, optimizes quilt
The boot sequence and corresponding restoration path for starting unit, make all kinds of power supplys start as early as possible, is grid-connected, enhance recovery system
Intensity provides power support for comprehensive recovery of subsequent load.Current unit recovery algorithms are mainly offline optimization method, i.e.,
Whole recovery scheme is formulated according to preset power failure scene and anticipation recovery process.In actual recovery process, after power failure
The state of system with when default scene is inconsistent or recovery process is not consistent with anticipation process, although the scheme formulated can be in advance
It gives management and running personnel certain guidance, but practical recovery process may be cannot be used directly for, influence unit recovering process.
Summary of the invention
The present invention to solve the above-mentioned problems, proposes a kind of online dynamic decision method and system of unit recovery.For
After having a power failure on a large scale in electric system original state and recovery process route recovery uncertainty, by deep learning and Meng Teka
Luo Shu search combines, the dynamic decision restored for unit.The present invention is searched by a variety of possible situations to subsequent recovery
Rope simulation improves robustness of its produced decision scheme when in face of uncertain situation, and can be according to power train
System real-time status, the transmission line of electricity that online dynamic decision will be put into next step are gradually completing the recovery of unit, various to cope with
Uncertain situation.
To achieve the goals above, the present invention adopts the following technical scheme:
The invention discloses a kind of units to restore online dynamic decision method, comprising the following steps:
(1) it determines black starting-up power supply when system is restored, and chooses the unit for needing preferentially to start from all units and make
For unit set to be restored;
(2) the tape label training set that covering unit as far as possible restores each possible state is generated;Using deep learning, for
Unit resumes training collection and is learnt, and constructs valuation network;
(3) it obtains unit and restores real time data, judge the scope of power outage of system, the state of system is identified, judge
The availability of various equipment in system obtains machine unit characteristic;
(4) valuation network is utilized, is search target with unit total generating capacity, Monte Carlo tree is carried out to each alternative route
Search;Meanwhile using the method for parallel computation, to voltage constraint, frequency constraint and the trend constraint after each alternative route investment
It is verified;
(5) summarize Monte Carlo tree search result, route choosing investment will be restored to next step.
Further, described according to system actual conditions, choose hydroelectric power plant, Pumped Storage Plant or gas turbine conduct
Black starting-up power supply.
Further, the unit for needing preferentially to start of choosing from all units is as unit set to be restored, choosing
The basic principle taken are as follows:
1) unit capacity to be restored is selected as 300MW to 600MW;
2) power plant unit capacity where unit is big;
3) preferentially start the unit near important load.
Further, the system data according to acquisition generates covering unit as far as possible and restores each possible state
Tape label training set, method particularly includes:
1) all possible shutdown status combination of unit is generated in a manner of traversal, it is assumed that N platform machine is shared in a certain system
The compressor emergency shutdown state of group, required generation is totalIt is a;
2) the number L of the lower required different line status combination generated of a certain set state combination is setnumWith a certain route
The number D of the different unit downtimes combination of required generation under combinations of statesnum;
3) line status is generated at random, and verifies the random topology connectivity for generating rack, it is desirable that all routes can connect
It connects and has restored unit and black starting-up power supply, line status adjustment is carried out to unqualified rack;
4) all anticipation recovery times for having restored route are calculated, and generate the downtime of every unit at random;
5) particle swarm optimization algorithm is used, optimization aim is up to unit total generating capacity, optimal unit is sought and restores
Scheme, and corresponding decision index system value is calculated, restore sample label as unit.
Further, the valuation network is the deep neural network after training;Establish the base containing 3 hidden layers
In the deep neural network training gained sample of sparse autocoder, input as Unit Commitment state, line status and unit
Downtime exports as unit total generating capacity optimal value.
Further, the state to system identifies, specifically includes: right after previous step transmission line of electricity investment
All grid equipments in newest scope of power outage carry out availability diagnosis, identify set available in recovery in next step
The specific downtime of standby and each fired power generating unit.
Further, it is search target with unit total generating capacity, the search of Monte Carlo tree, tool is carried out to each alternative route
Body method are as follows:
1) it selects: by root node, after calculating the improved upper limit confidence interval index value of each node, successively selecting
The maximum node of capping confidence interval index value carries out the extension or simulation of next step;
2) it extends: the number of child node is reduced using branch pruning technique, using branch pruning technique along newest node
Reverse search finds each node in each layer with identical father node, and going out for these nodes is avoided in newest extension
It is existing, until having new unit access;
3) it simulates: according to system mode, constantly subsequent optimizing decision index value quickly being estimated using valuation network
It calculates, and improves the probability that there is the corresponding alternative route of higher decision index system value to be selected, guide simulation process;
4) recall: after the completion of simulation, each node parameter in tree reversely being updated.
Further, p% before the improved upper limit confidence interval index value of each node sorts according to node index value
The number that the accessed number of the average value of the analog result parent of node node and the node are accessed determines.
Further, in the step (5), maximizing index using weighting unit generation ability will be extensive for next step
Multiple route choosing investment, specifically:
The decision function value of m-th of alternative measure be equal to m-th of alternative measure analog result each time index value with
M-th of alternative measure in simulation each time the ratio of route investment number cumulative and.
The invention discloses a kind of units to restore online dynamic decision system, comprising:
The device of black starting-up power supply when for determining that system is restored;
For choosing device of the unit for needing preferentially to start as unit set to be restored from all units;
The device for restoring the tape label training set of each possible state for generating covering unit as far as possible;
For constructing the device of valuation network;
Restore real time data for obtaining unit, judges the scope of power outage of system, the state of system is identified, judge
The availability of various equipment in system, obtains the device of machine unit characteristic;
It is search target with unit total generating capacity for utilizing valuation network, Monte Carlo is carried out to each alternative route
Set the device of search;
For carrying out the dress of parallel check to voltage constraint, frequency constraint and the trend constraint after each alternative route investment
It sets;
For summarizing Monte Carlo tree search result, the device of route choosing investment will be restored to next step.
The invention has the benefit that
(1) unit searched for based on deep learning and Monte Carlo tree that the present invention designs restores online dynamic decision side
Method, core innovative point are that real time data can be restored according to system, need to put into online dynamic decision unit recovery process
Transmission line of electricity, instruct the unit of power system recovery early period to restore;
(2) unit searched for based on deep learning and Monte Carlo tree that the present invention designs restores online dynamic decision side
Method, the unit that can preferably adapt to not homologous ray restore, and have stronger adaptability;
(3) unit searched for based on deep learning and Monte Carlo tree that the present invention designs restores online dynamic decision side
Method is completed the offline generation for restoring data for unit before unit restores on-line decision, and is carried out using deep learning
Data study, is applied in line process, saves the online dynamic decision time, improve the efficiency of decision-making;
(4) unit searched for based on deep learning and Monte Carlo tree that the present invention designs restores online dynamic decision side
Method is simulated by the search of a variety of possible situations to subsequent recovery, improves its produced decision scheme in face of uncertain
Robustness when character condition;
(5) unit searched for based on deep learning and Monte Carlo tree that the present invention designs restores online dynamic decision side
Method can quickly make a response when facing the case where not being consistent with anticipation process, provide unit recovery scheme.
Detailed description of the invention
Fig. 1 resumes training collection for unit and automatically generates flow chart;
Fig. 2 is deep neural network structure chart, restores the study of data for unit;
Fig. 3 is the unit recovery process modeling process chart based on valuation network;
Fig. 4 is the unit recovery measure search routine figure searched for based on deep learning and Monte Carlo tree;
Fig. 5 is that the unit searched for based on deep learning and Monte Carlo tree restores online dynamic decision method flow diagram;
Fig. 6 is Shandong West Region configuration of power network;
Fig. 7 is the Monte Carlo on-line search process for Shandong West Region power grid, and abscissa represents time, ordinate
It represents alternative route and is searched number proportion in entire search process.
Specific embodiment:
The invention will be further described with embodiment with reference to the accompanying drawing.
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
The invention discloses a kind of units to restore online dynamic decision method, and specific steps include:
(1) electric power system data is obtained.It determines black starting-up power supply when system is restored, and chooses and need from all units
The unit preferentially started is as unit set to be restored;
(2) according to system data, the tape label training set that covering unit as far as possible restores each possible state is generated.It utilizes
Deep learning resumes training collection for unit and learns, and constructs valuation network;
(3) it obtains unit and restores real time data.The scope of power outage of judgement system identifies the state of system, judges
The availability of various equipment in system obtains machine unit characteristic;
(4) valuation network is utilized, is search target with unit total generating capacity, Monte Carlo tree is carried out to each alternative route
Search.Meanwhile using the method for parallel computation, constraint checking is carried out to each alternative route;
(5) summarize Monte Carlo tree search result, choose the transmission line of electricity that will restore in next step, put by dispatcher.
In the step (1), system applied by the restoration methods is determined, and obtain system corresponding data.According to being
System situation chooses hydroelectric power plant, Pumped Storage Plant or gas turbine as black starting-up power supply.In recovery process, unit success
The frequency modulation and pressure regulation task of system are undertaken after starting, so unit capacity should not be too small, while in order to avoid set auxiliary machinery opens
Big impact is generated to system when dynamic, unit capacity to be restored is preferably selected as 300 to 600MW.In addition power plant's dress where unit
Machine capacity is big, and preferentially to start the unit near important load.By decision-maker's foundation mentioned above principle from all units
The unit for needing preferentially to start is selected as unit set to be restored.
In the step (2), the offline preparation of training set automatically generated before restoring online dynamic decision for unit.
Power system recovery state pole in practical power systems is not common, is difficult to find enough unit recovery samples in practice,
Related unit restores data needs and automatically generates offline.The data of each sample mainly include Unit Commitment state, line status
And unit downtime.Specific training set generation step is as follows:
1, all possible shutdown status combination of unit is generated in a manner of traversal, it is assumed that N platform machine is shared in a certain system
The compressor emergency shutdown state of group, required generation is totalIt is a;
2, the number of the different line status combination generated needed for a certain set state of parameter combines down is set by policymaker
LnumThe number D of the different unit downtimes combination generated needed for lower with the combination of a certain line statusnum, value is bigger, then instructs
The coverage area for practicing collection is wider, but the corresponding generation time will increase;
3, line status is generated at random, and verifies the random topology connectivity for generating rack, it is desirable that all routes can connect
It connects and has restored unit and black starting-up power supply, line status adjustment is carried out to unqualified rack;
4, all anticipation recovery times for having restored route are calculated, and generate the downtime of every unit at random.At certain
Under one rack form, unit downtime minimum 0 is up to and restores the time needed for whole routes in the rack.It considers
The uncertainty of route recovery time, after setting for 0 to anticipation recovery time for the downtime range for not restoring unit
10min;
5, state is restored according to the unit of each sample using particle swarm optimization algorithm, is up to unit total generating capacity
Optimization aim acquires subsequent optimal unit recovery scheme, and calculates corresponding decision index system value, as sample label.
Training set to automatically generate flow chart as shown in Figure 1.
Unit total generating capacity index is shown below:
Wherein, EtotalFor all units to be restored in recovery time T maximum generation ability, nGFor unit number to be restored
Mesh, Pi(t) go out force function for unit i to be restored
In the step (2), valuation network is the deep neural network after training, the simulation part being mainly used in MCTS
Point, state can be restored according to unit, quickly estimate the optimal value of decision index system.Establish the base containing 3 hidden layers
In the deep neural network training gained sample of sparse autocoder, input as Unit Commitment state, line status and unit
Downtime, exports as unit total generating capacity optimal value, and structure chart is as shown in Figure 2.
In the step (3), the state recognition of system is referred to, after previous step transmission line of electricity investment, to newest power failure
All grid equipments in range carry out availability diagnosis, identify in next step the available equipment in recovery, and each
The specific downtime of fired power generating unit.
It is search target with unit total generating capacity, and to the voltage constraint after route investment, frequency in the step (4)
Rate constraint and trend constraint carry out parallel check, and expression formula is as follows:
Wherein, EtotalFor all units to be restored in recovery time T maximum generation ability, nGFor unit number to be restored
Mesh, Pi(t) go out force function, U for unit i to be restoredkFor the voltage of node k,WithRespectively indicate node k voltage
Bound, f are the frequency of system, fminAnd fmaxThe respectively bound of system frequency, PlFor route l transmission effective power flow,For the effective power flow upper limit of route l.It, will in view of the rack of system initial stages of restoration is more fragileIt is set as being less than static state
One value of stability limit and the thermostabilization limit.
In the step (4), the search of Monte Carlo tree is the algorithm that optimizing decision is made in a kind of artificial intelligence problem,
Iterative process is broadly divided into four steps: selection, extension, simulation, backtracking.It is searched for using Monte Carlo tree and carries out unit recovery online certainly
Steps are as follows for plan:
1, it selects.By root node, after calculating improved upper limit confidence interval (MUCT) index value of each node,
Successively choose extension or simulation that the maximum node of MUCT index value carries out next step.MUCT Index Formula is shown below
Wherein, FMUCTTo improve UCT index,Indicate the average value of preceding 70% analog result of node c index value sequence, n
For the number that node c parent node is accessed, ncFor the number that node c is accessed, CpFor the real number greater than 0.
2, it extends.When extension, the number of child node is reduced using branch pruning technique, can increase algorithm search depth
And range, promote search efficiency.For branch pruning technique along newest node reverse search, finding has identical father node in each layer
Each node, and the appearance of these states is avoided in newest extension, until having new unit access.
3, it simulates.Constantly subsequent optimizing decision index value is quickly estimated according to system mode using valuation network
It calculates, and improves the probability that there is the corresponding alternative route of higher decision index system value to be selected, guide simulation process.Physical simulation stream
Journey figure is as shown in Figure 3.
4, recall.After the completion of simulation, each node parameter in tree is reversely updated.
The Monte Carlo tree search routine figure restored for unit is as shown in Figure 4.
In the step (5), Monte Carlo tree search result is summarized, and maximum using weighting unit generation ability
Route choosing investment will be restored for next step by changing index, and expression formula is shown below
Wherein, fmIndicate the decision function value of m-th of alternative measure,It is m-th of alternative measure in n-th simulation
Route puts into number,For the index value of the n-th analog result of m-th of alternative measure.
The present invention further discloses a kind of units to restore online dynamic decision system, comprising:
The device of black starting-up power supply when for determining that system is restored;
For choosing device of the unit for needing preferentially to start as unit set to be restored from all units;
The device for restoring the tape label training set of each possible state for generating covering unit as far as possible;
For constructing the device of valuation network;
Restore real time data for obtaining unit, judges the scope of power outage of system, the state of system is identified, judge
The availability of various equipment in system, obtains the device of machine unit characteristic;
It is search target with unit total generating capacity for utilizing valuation network, Monte Carlo is carried out to each alternative route
Set the device of search;
For carrying out the dress of parallel check to voltage constraint, frequency constraint and the trend constraint after each alternative route investment
It sets;
For summarizing Monte Carlo tree search result, the device of route choosing investment will be restored to next step.
It is emulated below for Shandong West Region power grid real system, illustrates that unit restores online dynamic decision method
Process.
Shandong West Region electric network composition is as shown in Figure 6, it is assumed that and a certain moment Pump of Zhou County Power Plant ', Heze Plant have been started up,
The downtime of remaining power plant is 35min, restores online dynamic decision method using unit and carries out gradually unit recovery, specifically
Steps are as follows:
S1: obtaining associate power system data, selects a black starting-up power supply reliable for operation, and selects capacity suitable
When place power plant unit capacity is big, and the high unit of perimeter load importance is as unit set to be restored.
In recovery process, unit will undertake system frequency modulation and pressure regulation task after successfully starting up is so unit capacity is unsuitable
It is too small, while in order to avoid generating big impact to system when set auxiliary machinery starting, therefore unit capacity to be restored is selected as
Between 300MW-600MW.Power plant unit capacity is big where unit simultaneously, and preferentially to start the machine near important load
Group.The unit for needing preferentially to start is selected from all power loss units according to mentioned above principle by decision-maker.
Black starting-up power supply first choice is Taishan Pumped Storage Power Station in Shandong Power, and wherein #1 unit is transformed by multiple, and
And 3 black startup tests were successively carried out, and it is safe and reliable to operation, it is ideal black starting-up power supply.It is preferred according to unit to be restored
Principle is chosen in Shi Heng second factory, Pump of Zhou County Power Plant ', canal power plant, Heze Plant, Huang Tai power plant, Liaocheng Thermal Power Plant and Hua De power plant
First unit to be launched is as unit set to be restored.
S2: according to system data, the offline unit that generates restores sample, and calculates sample mark using particle swarm optimization algorithm
Label.Sample data includes Unit Commitment state, line status and unit downtime.Part work is offline beam worker
Make, need to generate number of samples altogether is 630,000, about one week time-consuming.Generate the specific steps that unit restores sample are as follows:
1, all possible shutdown status combination of unit is generated in a manner of traversal, shares 7 in the power grid of Shandong West Region
Platform unit to be launched, the compressor emergency shutdown state of required generation totally 126;
2, parameter L is setnumAnd DnumRespectively 100 and 50.Wherein, LnumIndicate raw needed for a certain set state combination is lower
At different line status combination number, DnumWhen indicating the different compressor emergency shutdowns generated needed for a certain line status combination is lower
Between the number that combines;
3, line status is generated at random, and verifies the random topology connectivity for generating rack, it is desirable that all routes can connect
It connects and has restored unit and black starting-up power supply, line status adjustment is carried out to unqualified rack.
4, all anticipation recovery times for having restored route are calculated, and generate the downtime of every unit at random.Due to
The uncertainty of grid structure after power failure, under a certain rack form, unit downtime minimum 0 is up to and restores the net
Time needed for whole routes in frame.Accordingly, it is considered to the uncertainty of route recovery time be arrived, when by the shutdown for not restoring unit
Between range be set as 0 to anticipation recovery time after 10min.
5, state is restored according to the unit of each sample using particle swarm optimization algorithm, is up to unit total generating capacity
Optimization aim acquires subsequent optimal unit recovery scheme, and calculates corresponding decision index system value, as sample label.
S3: establishing the deep neural network containing 3 hidden layers, and the sample generated using S2, with discrete automatic
Based on encoder, neural network is trained, constructs valuation network.Part work is offline preparation, nerve net
The structure of network is set as [22910050201], and final gained network test error is 3.5%.
S4: obtaining system real-time recovery data, judge the scope of power outage of system, identify to system mode, and judgement is each
The availability of equipment obtains unit downtime.
According to system real time data, at a certain moment, Taishan Pumped Storage Power Station is had been turned in the power grid of Shandong West Region, Zou
Safe line, county in Shandong Province Thailand line, county in Shandong Province waterside line and He waterside line have been put into, and Pump of Zhou County Power Plant ' is grid-connected with Heze Plant.Each route in system
Investment, and its anticipation recovery time is identical as anticipation recovery time under normal condition, the downtime of remaining unit to be launched is equal
For 35min.
S5: carrying out going deep into search using the alternative route that valuation network and the search of Monte Carlo tree restore next step,
And voltage, frequency and trend constraint are verified.
In the step with unit total generating capacity be search target, and to route investment after voltage constraint, frequency constraint and
Trend constraint carries out parallel check, and expression formula is as follows:
Wherein, EtotalFor all units to be restored in recovery time T maximum generation ability, nGFor unit number to be restored
Mesh, Pi(t) go out force function, U for unit i to be restoredkFor the voltage of node k,WithRespectively indicate node k voltage
Bound, f are the frequency of system, fminAnd fmaxThe respectively bound of system frequency, PlFor route l transmission effective power flow,For the effective power flow upper limit of route l.It, will in view of the rack of system initial stages of restoration is more fragileIt is set as being less than static state
One value of stability limit and the thermostabilization limit.
Searching for progress unit recovery on-line decision using Monte Carlo tree, steps are as follows:
1, it selects.By root node, after calculating improved upper limit confidence interval (MUCT) index value of each node,
Successively choose extension or simulation that the maximum node of MUCT index value carries out next step.MUCT Index Formula is shown below
Wherein, FMUCTTo improve UCT index,Indicate the average value of preceding 70% analog result of node c index value sequence, n
For the number that node c parent node is accessed, ncFor the number that node c is accessed, CpFor the real number greater than 0.
2, it extends.When extension, the number of child node is reduced using branch pruning technique, can increase algorithm search depth
And range, promote search efficiency.For branch pruning technique along newest node reverse search, finding has identical father node in each layer
Each node, and the appearance of these states is avoided in newest extension, until having new unit access.
3, it simulates.Constantly subsequent optimizing decision index value is quickly estimated according to system mode using valuation network
It calculates, and improves the probability that there is the corresponding alternative route of higher decision index system value to be selected, guide simulation process.
4, recall.After the completion of simulation, each node parameter in tree is reversely updated.
320s (recovery time of previous route is 6min) is set by the search time of MCTS, final search result is such as
Shown in Fig. 7, with the progress of search, each route is gradually distinguished, and final selected route " safe antenna " is opened from 100s
The searched number proportion that begins just is higher than remaining route, and gap is increasing.
S6: summarizing Monte Carlo tree search result, chooses the transmission line of electricity that will restore in next step, is put by dispatcher.
It in the step, chooses " safe antenna " and is put by dispatcher, and return step S4, continue to put into route in next step
Selection, until completing the recovery of all units.
Final gained unit recovery sequence are as follows: "-Huang Tai power plant-Hua De power plant, Shi Heng second power plant-Liaocheng thermoelectricity-canal electricity
Factory." corresponding route recovery sequence are as follows: " the splendid line-Shao Linxian-of safe antenna-day garden line-stone garden line-splendid line-Huang of Jinan-Tai'an line-Ji is magnificent
Face line-and chats splendid line-merely online-upper river line in high mountain line-Xu Yue line-county in Shandong Province."
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.
Claims (9)
1. a kind of unit restores online dynamic decision method, which comprises the following steps:
(1) determine black starting-up power supply when system is restored, and choose from all units the unit that needs preferentially to start be used as to
Restore unit set;
(2) the tape label training set that covering unit as far as possible restores each possible state is generated;Using deep learning, for unit
It resumes training collection to be learnt, constructs valuation network;
(3) it obtains unit and restores real time data, judge the scope of power outage of system, the state of system is identified, judges system
In various equipment availability, obtain machine unit characteristic;
(4) valuation network is utilized, is search target with unit total generating capacity, Monte Carlo tree is carried out to each alternative route and is searched
Rope;Meanwhile using the method for parallel computation, to voltage constraint, frequency constraint and the trend constraint after each alternative route investment into
Row verification;The search of Monte Carlo tree is the algorithm that optimizing decision is made in a kind of artificial intelligence problem, and iterative process is mainly divided
Extension, simulation, recall for four steps: selection;Searching for progress unit recovery on-line decision using Monte Carlo tree, steps are as follows:
1, it selects;By root node, after calculating the improved upper limit confidence interval MUCT index value of each node, successively select
The maximum node of MUCT index value is taken to carry out the extension or simulation of next step;MUCT Index Formula is shown below
Wherein, FMUCTTo improve UCT index,Indicate that the average value of preceding 70% analog result of node c index value sequence, n are section
The accessed number of point c parent node, ncFor the number that node c is accessed, CpFor the real number greater than 0;
2, it extends;When extension, the number of child node is reduced using branch pruning technique, increases algorithm search depth and range,
Promote search efficiency;Branch pruning technique finds each section in each layer with identical father node along newest node reverse search
Point, and the appearance of these states is avoided in newest extension, until having new unit access;
3, it simulates;Constantly subsequent optimizing decision index value is quickly estimated according to system mode using valuation network,
And the probability that there is the corresponding alternative route of higher decision index system value to be selected is improved, guide simulation process;
4, recall;After the completion of simulation, each node parameter in tree is reversely updated;
(5) summarize Monte Carlo tree search result, route choosing investment will be restored to next step.
2. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that according to the practical feelings of system
Condition chooses hydroelectric power plant, Pumped Storage Plant or gas turbine as black starting-up power supply.
3. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that described from all units
The middle unit for needing preferentially to start of choosing is as unit set to be restored, the basic principle of selection are as follows:
1) unit capacity to be restored is selected as 300MW to 600MW;
2) power plant unit capacity where unit is big;
3) preferentially start the unit near important load.
4. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that according to the system of acquisition
Data generate the tape label training set that covering unit as far as possible restores each possible state, method particularly includes:
1) all possible shutdown status combination of unit is generated in a manner of traversal, it is assumed that N platform unit, institute are shared in a certain system
The compressor emergency shutdown state that need to be generated is totalIt is a;
2) the number L of the lower required different line status combination generated of a certain set state combination is setnumWith a certain line status
The number D of the different unit downtimes combination generated needed for combination is lowernum;
3) line status is generated at random, and verifies the random topology connectivity for generating rack, it is desirable that all routes can connect
Restore unit and black starting-up power supply, line status adjustment is carried out to unqualified rack;
4) all anticipation recovery times for having restored route are calculated, and generate the downtime of every unit at random;
5) particle swarm optimization algorithm is used, optimization aim is up to unit total generating capacity, seeks optimal unit recovery scheme,
And corresponding decision index system value is calculated, restore sample label as unit.
5. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that the valuation network is
Deep neural network after training;Establish the deep neural network based on sparse autocoder containing 3 hidden layers
Training gained sample, inputs as Unit Commitment state, line status and unit downtime, exports as unit total generating capacity most
The figure of merit.
6. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that the shape to system
State is identified, is specifically included: previous step transmission line of electricity investment after, to all grid equipments in newest scope of power outage into
The diagnosis of row availability, identifies the specific downtime of available equipment and each fired power generating unit in recovery in next step.
7. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that each node changes
Into upper limit confidence interval index value sorted according to node index value before the average value of the p% analog result parent of node node quilt
The number that the number of access and the node are accessed determines.
8. a kind of unit as described in claim 1 restores online dynamic decision method, which is characterized in that sent out using weighting unit
Electric energy power, which maximizes index, will restore route choosing investment for next step, specifically:
The decision function value of m-th of alternative measure is equal to the index value of the analog result each time of m-th of alternative measure and m-th
Alternative measure in simulation each time the ratio of route investment number cumulative and.
9. a kind of unit restores online dynamic decision system characterized by comprising
The device of black starting-up power supply when for determining that system is restored;
For choosing device of the unit for needing preferentially to start as unit set to be restored from all units;
The device for restoring the tape label training set of each possible state for generating covering unit as far as possible;
For constructing the device of valuation network;
Restore real time data for obtaining unit, judges the scope of power outage of system, the state of system is identified, judges system
In various equipment availability, obtain the device of machine unit characteristic;
For utilizing valuation network, it is search target with unit total generating capacity, Monte Carlo tree is carried out to each alternative route and is searched
The device of rope;The search of Monte Carlo tree is the algorithm that optimizing decision is made in a kind of artificial intelligence problem, and iterative process is main
It is divided into four steps: selection, extension, simulation, backtracking;Searching for progress unit recovery on-line decision using Monte Carlo tree, steps are as follows:
1, it selects;By root node, after calculating the improved upper limit confidence interval MUCT index value of each node, successively select
The maximum node of MUCT index value is taken to carry out the extension or simulation of next step;MUCT Index Formula is shown below
Wherein, FMUCTTo improve UCT index,Indicate that the average value of preceding 70% analog result of node c index value sequence, n are section
The accessed number of point c parent node, ncFor the number that node c is accessed, CpFor the real number greater than 0;
2, it extends;When extension, the number of child node is reduced using branch pruning technique, increases algorithm search depth and range,
Promote search efficiency;Branch pruning technique finds each section in each layer with identical father node along newest node reverse search
Point, and the appearance of these states is avoided in newest extension, until having new unit access;
3, it simulates;Constantly subsequent optimizing decision index value is quickly estimated according to system mode using valuation network,
And the probability that there is the corresponding alternative route of higher decision index system value to be selected is improved, guide simulation process;
4, recall;After the completion of simulation, each node parameter in tree is reversely updated;
For carrying out the device of parallel check to voltage constraint, frequency constraint and the trend constraint after each alternative route investment;
For summarizing Monte Carlo tree search result, the device of route choosing investment will be restored to next step.
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810271593.3A CN108429259B (en) | 2018-03-29 | 2018-03-29 | A kind of online dynamic decision method and system of unit recovery |
US16/348,190 US11159018B2 (en) | 2018-03-29 | 2018-09-29 | Method and system for online decision making of generator start-up |
PCT/CN2018/108531 WO2019184286A1 (en) | 2018-03-29 | 2018-09-29 | Online dynamic decision-making method and system for unit restoration |
EP18900563.0A EP3780307B1 (en) | 2018-03-29 | 2018-09-29 | Online dynamic decision-making method and system for unit restoration |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810271593.3A CN108429259B (en) | 2018-03-29 | 2018-03-29 | A kind of online dynamic decision method and system of unit recovery |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108429259A CN108429259A (en) | 2018-08-21 |
CN108429259B true CN108429259B (en) | 2019-10-18 |
Family
ID=63160017
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810271593.3A Active CN108429259B (en) | 2018-03-29 | 2018-03-29 | A kind of online dynamic decision method and system of unit recovery |
Country Status (4)
Country | Link |
---|---|
US (1) | US11159018B2 (en) |
EP (1) | EP3780307B1 (en) |
CN (1) | CN108429259B (en) |
WO (1) | WO2019184286A1 (en) |
Families Citing this family (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108429259B (en) * | 2018-03-29 | 2019-10-18 | 山东大学 | A kind of online dynamic decision method and system of unit recovery |
CN111049176B (en) * | 2018-10-15 | 2023-10-24 | 云南电网有限责任公司怒江供电局 | Method for selecting black start network reconstruction optimal path |
US11593693B2 (en) * | 2019-01-23 | 2023-02-28 | Royal Bank Of Canada | System and method for tree-based machine learning |
CN109861297B (en) * | 2019-04-11 | 2022-11-01 | 上海电机学院 | Black start method of power system based on grey wolf optimization algorithm |
CN111047915B (en) * | 2019-12-13 | 2020-11-27 | 中国科学院深圳先进技术研究院 | Parking space allocation method and device and terminal equipment |
CN111401729A (en) * | 2020-03-12 | 2020-07-10 | 国网山东省电力公司潍坊供电公司 | Newly-increased black start unit distribution point determining method and system |
CN111428929B (en) * | 2020-03-23 | 2023-06-06 | 复旦大学 | Path planning system with two-dimensional loading and unloading and time window limitation |
CN111934948B (en) * | 2020-07-20 | 2022-04-12 | 浙江理工大学 | EFSM executable test sequence generation method based on Monte Carlo tree search |
EP4193443A1 (en) * | 2020-09-14 | 2023-06-14 | Siemens Corporation | Power distribution system reconfigurations for multiple contingencies |
CN112117761B (en) * | 2020-09-14 | 2021-12-28 | 西安交通大学 | New energy power system recovery path searching method based on improved network weight |
CN112288237A (en) * | 2020-10-12 | 2021-01-29 | 国网江苏省电力有限公司镇江供电分公司 | Black starter subregion partitioning method based on synergy optimization |
CN112348262A (en) * | 2020-11-10 | 2021-02-09 | 国网上海市电力公司 | Power cable operation safety early warning method |
CN112531741B (en) * | 2020-12-04 | 2023-03-24 | 广东电网有限责任公司电力调度控制中心 | Decision method and device for flexible load to participate in frequency modulation |
CN112668946B (en) * | 2021-01-28 | 2022-07-05 | 广西大学 | VSC-HVDC (Voltage Source converter-high Voltage direct Current) access power system unit recovery sequence decision method |
CN112765828B (en) * | 2021-01-28 | 2022-09-27 | 中国人民解放军海军工程大学 | Real-time simulation network subdivision method for complex power system |
CN112837739B (en) * | 2021-01-29 | 2022-12-02 | 西北大学 | Hierarchical feature phylogenetic model based on self-encoder and Monte Carlo tree |
CN112803470B (en) * | 2021-02-08 | 2023-01-20 | 华北电力大学(保定) | Alternating current grid reconstruction method for supporting starting of LCC-HVDC system |
CN113139644B (en) * | 2021-03-24 | 2024-02-09 | 北京科技大学顺德研究生院 | Information source navigation method and device based on deep Monte Carlo tree search |
US20220398283A1 (en) * | 2021-05-25 | 2022-12-15 | Nvidia Corporation | Method for fast and better tree search for reinforcement learning |
CN113612223B (en) * | 2021-08-05 | 2024-05-24 | 云南电力技术有限责任公司 | Hydropower station regional power grid black start modeling and simulation method |
CN113704098B (en) * | 2021-08-18 | 2023-09-22 | 武汉大学 | Deep learning fuzzy test method based on Monte Carlo search tree seed scheduling |
CN113988558B (en) * | 2021-10-20 | 2022-08-19 | 山东大学 | Power grid dynamic security assessment method based on blind area identification and electric coordinate system expansion |
CN114256886B (en) * | 2021-12-22 | 2023-08-08 | 合肥工业大学 | Confidence interval decision-based unit recovery sequence optimization method |
CN114330119B (en) * | 2021-12-27 | 2023-08-25 | 淮阴工学院 | Deep learning-based extraction and storage unit adjusting system identification method |
CN114583705B (en) * | 2022-03-29 | 2024-03-12 | 合肥工业大学 | Black-start power supply planning method considering power flow of power system |
CN115118477B (en) * | 2022-06-22 | 2024-05-24 | 四川数字经济产业发展研究院 | Smart grid state recovery method and system based on deep reinforcement learning |
CN115936323A (en) * | 2022-08-15 | 2023-04-07 | 苏州诀智科技有限公司 | Intelligent in-cabin stowage method and system based on container ship and computer medium |
CN117877608B (en) * | 2024-03-13 | 2024-05-28 | 烟台国工智能科技有限公司 | Monte Carlo tree search inverse synthesis planning method and device based on experience network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101236575A (en) * | 2008-01-30 | 2008-08-06 | 山东大学 | Black start assistant decision making support/training system and its method |
WO2009112955A2 (en) * | 2008-03-10 | 2009-09-17 | Eomax Corporation | Modular imaging system with wireless transmission and locking connector |
CN103248127A (en) * | 2013-05-23 | 2013-08-14 | 山东大学 | Multi-space-time navigating power system restoration decision support system and restoration decision method |
CN103279620A (en) * | 2013-06-07 | 2013-09-04 | 山东大学 | Method for restoring sequence and path of unit and simultaneously performing optimization |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9639904B2 (en) * | 2012-12-11 | 2017-05-02 | Opterra Energy Services, Inc. | Systems and methods for minimizing energy costs for a power consumption system that has access to off-grid resources |
US9602021B2 (en) * | 2014-03-07 | 2017-03-21 | General Electric Company | Hybrid high voltage direct current converter system and method of operating the same |
US20160139212A1 (en) * | 2014-11-13 | 2016-05-19 | Eleon Energy, Inc. | Methods and systems for power restoration planning employing simulation and transient test analysis |
CN107244312B (en) * | 2017-05-23 | 2019-11-26 | 上海蔚来汽车有限公司 | Electrical changing station and orderly change controller for electric consumption, method |
CN108429259B (en) * | 2018-03-29 | 2019-10-18 | 山东大学 | A kind of online dynamic decision method and system of unit recovery |
-
2018
- 2018-03-29 CN CN201810271593.3A patent/CN108429259B/en active Active
- 2018-09-29 WO PCT/CN2018/108531 patent/WO2019184286A1/en active Application Filing
- 2018-09-29 US US16/348,190 patent/US11159018B2/en active Active
- 2018-09-29 EP EP18900563.0A patent/EP3780307B1/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101236575A (en) * | 2008-01-30 | 2008-08-06 | 山东大学 | Black start assistant decision making support/training system and its method |
WO2009112955A2 (en) * | 2008-03-10 | 2009-09-17 | Eomax Corporation | Modular imaging system with wireless transmission and locking connector |
CN103248127A (en) * | 2013-05-23 | 2013-08-14 | 山东大学 | Multi-space-time navigating power system restoration decision support system and restoration decision method |
CN103279620A (en) * | 2013-06-07 | 2013-09-04 | 山东大学 | Method for restoring sequence and path of unit and simultaneously performing optimization |
Non-Patent Citations (1)
Title |
---|
考虑机组恢复成功率的黑启动阶段待恢复机组优选;朱海南 等;《电力系统自动化》;20131125;第37卷(第22期);第28-34段 * |
Also Published As
Publication number | Publication date |
---|---|
US11159018B2 (en) | 2021-10-26 |
WO2019184286A1 (en) | 2019-10-03 |
EP3780307A1 (en) | 2021-02-17 |
CN108429259A (en) | 2018-08-21 |
EP3780307B1 (en) | 2023-03-08 |
US20200127457A1 (en) | 2020-04-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108429259B (en) | A kind of online dynamic decision method and system of unit recovery | |
Guo et al. | A multi-state model for transmission system resilience enhancement against short-circuit faults caused by extreme weather events | |
CN103812131B (en) | A kind of urban distribution network isolated island black starting-up system and method based on multiple agent | |
CN103400302B (en) | A kind of wind power base cascading failure risk perceptions method for early warning | |
Miller | Keeping it together: Transient stability in a world of wind and solar generation | |
CN103904682B (en) | A kind of power forecasting method based on scene mixed model | |
Liu et al. | Research on fault scenario prediction and resilience enhancement strategy of active distribution network under ice disaster | |
CN112613685B (en) | Disaster power distribution network rush-repair method considering dynamic influence of external environment | |
CN110336270A (en) | A kind of update method of Power system transient stability prediction model | |
CN115310378A (en) | Power grid toughness evaluation and differentiation planning method under extreme typhoon disaster | |
CN104868465A (en) | Power system grid structure reconfiguration and optimization method based on fuzzy chance constraint | |
CN103107544A (en) | On-line pre-deciding method for offshore oilfield group power grid emergency control | |
CN110391657B (en) | Method for improving toughness of power system for dealing with short-circuit fault caused by extreme weather | |
CN107679723B (en) | Networked remote testing method for new energy power generation grid-connected system | |
CN111027855A (en) | Power system risk control method considering power transmission line meteorological disaster fault probability | |
CN112688313A (en) | Active power distribution network reliability assessment method based on matrix modeling | |
Pakdel | Intelligent instability detection for islanding prediction | |
CN104269846A (en) | Power grid fault partitioning method | |
CN105958474A (en) | Power transmission line dynamic capacity increasing method and system used for power grid regulation and control system | |
CN108183481A (en) | One kind quickly sentences steady method and system based on deep learning power grid | |
CN103065193A (en) | On-line intelligent identification method of provincial level power grid cascading failures | |
CN104281878A (en) | Electric power system black start on-line navigation type decision recovery method | |
CN115906610A (en) | Distributed power supply site selection planning method considering line faults and power grid toughness | |
CN109525437A (en) | Distribution Network Communication node restoration methods and system | |
Wang et al. | Early warning of distribution transformer based on bp neural network considering the influence of extreme weather |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |